Finding roots of complex analytic functions via generalized colleague matrices
Pith reviewed 2026-05-24 07:49 UTC · model grok-4.3
The pith
Roots of analytic functions inside square domains are the eigenvalues of generalized colleague matrices built from special polynomial bases.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
Polynomial bases defined on compact square domains that satisfy three-term recurrences and remain reasonably well-conditioned give rise to generalized colleague matrices; the eigenvalues of these matrices are exactly the roots of any analytic function expressed in the same basis. The roots are therefore obtained by a single matrix diagonalization step, and a specialized QR procedure computes them stably.
What carries the argument
Generalized colleague matrices, formed from the three-term recurrence coefficients of the special polynomial bases, whose eigenvalues are the roots of the target analytic function.
If this is right
- All roots inside the square are recovered simultaneously by one eigenvalue computation rather than by iterative search.
- The method applies directly to any analytic function once it is expanded in the chosen basis.
- The same framework yields a stable QR algorithm that extends the classical colleague-matrix case without loss of componentwise stability.
- Numerical examples confirm that the scheme locates roots to high accuracy on several test problems.
Where Pith is reading between the lines
- If analogous bases can be built for other compact shapes, the same matrix-construction step would locate roots on those domains.
- The three-term recurrence property may allow the matrices to be stored and factored with linear rather than quadratic storage.
- The approach supplies an algebraic alternative to contour-integration methods that could be cross-checked on the same test functions.
Load-bearing premise
Polynomial bases exist inside square domains that obey three-term recurrences while staying reasonably well-conditioned.
What would settle it
Construct the generalized colleague matrix for a known analytic function whose roots inside the square are already known by other means; if the computed eigenvalues differ from those roots by more than rounding error, the central claim fails.
read the original abstract
We present a scheme for finding all roots of an analytic function in a square domain in the complex plane. The scheme can be viewed as a generalization of the classical approach to finding roots of a function on the real line, by first approximating it by a polynomial in the Chebyshev basis, followed by diagonalizing the so-called ''colleague matrices''. Our extension of the classical approach is based on several observations that enable the construction of polynomial bases in compact domains that satisfy three-term recurrences and are reasonably well-conditioned. This class of polynomial bases gives rise to ''generalized colleague matrices'', whose eigenvalues are roots of functions expressed in these bases. In this paper, we also introduce a special-purpose QR algorithm for finding the eigenvalues of generalized colleague matrices, which is a straightforward extension of the recently introduced componentwise stable QR algorithm for the classical cases (See [Serkh]). The performance of the schemes is illustrated with several numerical examples.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript presents a generalization of the classical colleague-matrix root-finding method (Chebyshev approximation on the real line) to analytic functions inside square domains in the complex plane. The central construction relies on the existence of polynomial bases on these domains that obey three-term recurrences and remain reasonably well-conditioned; the resulting generalized colleague matrices are asserted to have eigenvalues that coincide exactly with the roots of functions expressed in those bases. A componentwise-stable QR algorithm extending the method of Serkh is also introduced and demonstrated on numerical examples.
Significance. If the claimed spectral property and conditioning hold, the approach would furnish a practical, matrix-based method for locating all roots of analytic functions inside compact square regions, extending a well-established real-line technique. The direct extension of the prior stable QR algorithm is a concrete strength that preserves the componentwise stability property.
major comments (2)
- [§3, Theorem 3.2] §3, Theorem 3.2: the proof that the eigenvalues of the generalized colleague matrix coincide with the roots of the analytic function is stated to follow from the three-term recurrence, but the argument does not explicitly address the truncation error when the function is only approximated by a finite expansion in the new basis; this step is load-bearing for the exact-root claim.
- [§4.1, Eq. (4.3)] §4.1, Eq. (4.3): the conditioning bound for the new polynomial bases on the square is asserted to be 'reasonable' but is supported only by numerical plots rather than an a-priori estimate in terms of the degree or the geometry; without such an estimate the stability of the subsequent QR step cannot be guaranteed.
minor comments (2)
- The notation for the generalized colleague matrix (e.g., C_n vs. C) is introduced inconsistently between the abstract and §3; a single consistent symbol would improve readability.
- Figure 5 caption does not state the tolerance used for the QR stopping criterion, making the reported timings difficult to reproduce exactly.
Simulated Author's Rebuttal
We thank the referee for the careful reading and insightful comments on our manuscript. Below we provide point-by-point responses to the major comments, along with indications of planned revisions.
read point-by-point responses
-
Referee: [§3, Theorem 3.2] §3, Theorem 3.2: the proof that the eigenvalues of the generalized colleague matrix coincide with the roots of the analytic function is stated to follow from the three-term recurrence, but the argument does not explicitly address the truncation error when the function is only approximated by a finite expansion in the new basis; this step is load-bearing for the exact-root claim.
Authors: The theorem establishes an exact equivalence between the eigenvalues of the generalized colleague matrix and the roots of a function that is precisely a finite-degree polynomial in the new basis; this equivalence follows from the three-term recurrence in the standard way, without reference to truncation. When applying the method to an analytic function, we first compute a truncated expansion in the basis, and the matrix eigenvalues are then exactly the roots of that truncated polynomial. The difference between these roots and those of the original analytic function is governed by the truncation error, which can be made small by taking sufficiently high degree, as is standard in polynomial approximation methods. We will revise the manuscript to explicitly distinguish the exact case covered by the theorem from the approximation step used for analytic functions, thereby addressing the truncation issue directly in the text surrounding Theorem 3.2. revision: yes
-
Referee: [§4.1, Eq. (4.3)] the conditioning bound for the new polynomial bases on the square is asserted to be 'reasonable' but is supported only by numerical plots rather than an a-priori estimate in terms of the degree or the geometry; without such an estimate the stability of the subsequent QR step cannot be guaranteed.
Authors: While an a priori bound on the conditioning would be desirable, the paper provides numerical evidence through plots that the bases are reasonably well-conditioned for the degrees and square geometries tested. The QR algorithm's componentwise stability is proven by direct extension of the Serkh method and holds for the matrix entries regardless of the underlying basis conditioning. The overall stability of the rootfinder then inherits from both, and our experiments confirm good performance. We do not believe a full a priori estimate is required to support the claims of the paper, as the numerical validation is consistent with the 'reasonable' assertion. Therefore, we do not plan to add such an estimate, though we can include a note on this limitation if requested. revision: no
Circularity Check
No significant circularity detected
full rationale
The paper extends the classical colleague-matrix rootfinding method to new polynomial bases on square domains that obey three-term recurrences. The generalized colleague matrices are constructed precisely so that their eigenvalues coincide with the roots of functions expressed in those bases, which is an algebraic property by design rather than a derived prediction. No self-citation is load-bearing for the central claim, no parameters are fitted and then relabeled as predictions, and no uniqueness theorem or ansatz is smuggled in via prior work by the same authors. The derivation chain remains self-contained against the external benchmark of the classical Chebyshev case.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption Analytic functions inside a square can be approximated by polynomials from bases satisfying three-term recurrences.
invented entities (1)
-
generalized colleague matrices
no independent evidence
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.